Prediction of perception: Probing the hOR17-4 olfactory receptor model with silicon analogues of bourgeonal and lilial

Article abstract of DOI:10.1002/anie.200605002

(Figure Presented) Sense and sensibility: Silicon analogues of the lily-of-the-valley odorants lilial and bourgeonal demonstrate that the electronic surface structure determines the interaction of an odorant with its olfactory receptor. The subtle changes

Full text of DOI:10.1002/anie.200605002

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Angewandte
Chemie
DOI: 10.1002/anie.200605002
Carbon–Silicon Exchange
Prediction of Perception: Probing the hOR17-4 Olfactory Receptor
Model with Silicon Analogues of Bourgeonal and Lilial**
Leszek Doszczak, Philip Kraft,* Hans-Peter Weber, Rüdiger Bertermann, Annika Triller,
Hanns Hatt,* and Reinhold Tacke*
The award of the Nobel prize to Axel[1a] and Buck[1b] for their
pioneering studies on the mechanism of odor perception
generated high expectations concerning the prediction of
olfactory properties. This was perhaps unjustified, as Buck
et al.[2a] also discovered that each odorant triggers a combi-
nation of different receptors, that odors consequently corre-
spond to complex activation patterns of glomeruli, and that
mixtures can even stimulate cortical neurons that are not
stimulated by their individual components.[2b] Thus, it seems
almost impossible to predict the olfactory properties of a new
odorant that has a different affinity to some estimated 347
different human olfactory receptors.[3] Consequently, the
design of new odorants relies heavily on structural similarities
to the reference compound rather than on the complemen-
tarity to the receptor binding site.[4,5] To what extent, however,
can we predict the odor and intensity of a new compound on
the basis of an olfactory receptor model?
the odor, only the thresholds and nuances. Lyral[7a] and lilial
(1a)[7b] were the first two ligands for which odorant–receptor
complexes were computed by employing the OR5 receptor of
rats (= O1r1469) expressed in the baculovirus-Sf9 cell
system.[8] In the human olfactory epithelium, the hOR17-4
receptor (= OR1D2) responds to lilial (1a) and bourgeonal
(2a). It can also be used for quantitative evaluations as it is
also expressed in human sperm cells.[9] Furthermore, the
hOR17-40 receptor (= OR3A1) in the human olfactory
epithelium responds to lilial (1a),[10] but still the hOR17-4
receptor seemed ideal for a comparison of the in vivo, in vitro,
and in silico data.
To put our understanding of the odorant–receptor inter-
actions to the test we decided to predict the effect of the
substitution of a carbon atom for a silicon atom[6] in the lily-
of-the-valley odorants lilial (1a) and bourgeonal (2a). This
C/Si exchange (sila substitution) would affect their molecular
shape only subtly, which according to the latest lily-of-the-
valley olfactophore[5] should not affect the main character of
A homology model of the hOR17-4 receptor, based on the
crystal structure of bovine rhodopsin (1U19), was generated
with the MOLOC software.[11] The sequence alignment of
hOR17-4 to 1U19 was based on the published alignment of
OR1E1 to bovine rhodopsin,[12] which has a high homology to
hOR17-4. In the resulting model, the ligand binding pocket
(LBP) is lined with 21 amino acids of the transmembrane
(TM) helices plus two of the extracellular loop EL2, thereby
forming an almost closed cavity (see the Supporting Infor-
mation).
[*] Dr. L. Doszczak, Dr. R. Bertermann, Prof. Dr. R. Tacke
Universität Würzburg
As for rhodopsin, it is difficult to imagine how a ligand
should enter or leave the LBP. Possibly, the long, double-
folded EL2 chain, which seals the extracellular side, unfolds
and opens the entrance to the LBP. It has also been
suggested[13] that small lipophilic ligands could enter the
LBP sideways between TM5 and TM6 when the phenyl ring
of F212 (TM5) and the sidechain of I269 (TM6) move to
generate another rotamer.
Institut für Anorganische Chemie
Am Hubland, 97074 Würzburg (Germany)
Fax: (+49)931-888-4609
E-mail: r.tacke@mail.uni-wuerzburg.de
Dr. P. Kraft, Dr. H.-P. Weber
Givaudan Schweiz AG
Fragrance Research
Überlandstrasse 138, 8600 Dübendorf(Switzerland)
Fax: (+41)44-824-2926
Random, but stereochemically reasonable, conformations
of the respective odorants are flexibly fitted into the LBP,
whose 23 amino acids have flexible sidechains in an otherwise
rigid receptor, by using the MOLOC docking procedure
(Figure 1). The docked ligands were energetically optimized
to convergence, and ranked according to their enthalpic
binding energy. The best binding energies DEinter,corr for lilial
(1a) and bourgeonal (2a) and their silicon analogues 1b and
2b are summarized in Table 1. Lilial (1a) and its silicon
analogue 1b possess stereogenic centers, but racemize readily,
possibly also under physiological conditions. Therefore, we
studied the racemic mixtures and averaged the energy values
of the enantiomeric forms of 1a and 1b. There are contra-
dictory reports about the odor differences of the enantiomers
E-mail: philip.kraft@givaudan.com
A. Triller, Prof. Dr. Dr. Dr. H. Hatt
Lehrstuhl für Zellphysiologie
Ruhr-Universität Bochum
Universitätsstrasse 150, 44801 Bochum (Germany)
Fax: (+49)234-321-4129
E-mail: hanns.hatt@rub.de
[**] We would like to thank Alain E. Alchenberger (Givaudan) for the
olfactory evaluation and Katarina Grman (Givaudan) for determin-
ing the GC thresholds. P.K. is grateful to Boris Schilling and Stephan
Bieri (both Givaudan) for stimulating discussions on the bio-
chemistry of olfaction. L.D. is grateful to Givaudan Schweiz AG for
financial support.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the authors.
Angew. Chem. Int. Ed. 2007, 46, 3367 –3371
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3367

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Communications
indeed reflect the trends of the biological data obtained
(Table 1).
The synthesis of racemic sila-lilial (1b; Scheme 1) started
from 1,4-diiodobenzene (3), which upon monolithiation and
Scheme 1. Synthesis ofsila-lilial ( 1b).
treatment with chlorotrimethylsilane afforded (4-iodophe-
nyl)trimethylsilane (4). This was then converted into 1b by a
Heck reaction[15] with 2-methylprop-2-en-1-ol. An analogous
Heck reaction of 4 with prop-2-en-1-ol however furnished
only traces of the expected sila-bourgeonal (2b). Under the
reaction conditions applied, 2b immediately gave the aldol
condensation product 5 (Scheme 2, see the Supporting
Information). Thus, an alternative route via a dimethyl-
hydrazone[16] was chosen.
Figure 1. Docking ofA) ( S)-lilial [(S)-1a], B) (R)-lilial [(R)-1a], C) bour-
geonal (2a) (all shown in yellow) and their respective silicon analogues
(S)-1b, (R)-1b, and 2b (cyan) on the calculated hOR17-4 receptor
model, and D) superposition ofall ligands.
Table 1: Comparison ofcalculated binding energies DEinter,corr, measured
GC odor thresholds, and EC50 values for sperm cells.
hOR17-4 ligand
DEinter,corr
GC threshold EC50 (sperm cells)
[kcalmolÀ1
]
[ngLÀ1 air]
[mm]
lilial (1a)
sila-lilial (1b)
bourgeonal (2a)
sila-bourgeonal (2b)
À10.4
À5.5
0.10
3.30
0.16
0.55
10
20
10
15
À12.1
À7.5
Scheme 2. Reaction of 4 with prop-2-en-1-ol.
of lilial (1a), but the R enantiomer has always been found to
be the more intense one.[14] The differences were not too
pronounced in our calculations, but it is also the R enantiomer
that binds more strongly in both cases (see the Supporting
Information).
(4-Methylphenyl)trimethylsilane (7) was prepared from 4-
bromotoluene (6) according to Ref. [17] (Scheme 3). Radical
bromination of 7 with N-bromosuccinimide (NBS)[18] fur-
nished [4-(bromomethyl)phenyl]trimethylsilane (8), which
was treated with lithiated ethanal N,N-dimethylhydrazone
to provide the corresponding hydrazone 9, which upon
Figure 1 shows that all the ligands are bound to the
hOR17-4 model receptor in very similar positions. The formyl
groups of all ligands form two hydrogen bonds, although of
different strength, to both the NH2 moiety of N125 (TM3) and
the OH group of Y261 (TM6). Also, the strong van der Waals
interactions of their aromatic rings with the hydrophobic
clamp consisting of Y268 (TM6) and V121 (TM3) varies
distinctly in intensity (see the Supporting Information). The
silicon analogues 1b and 2b show consistently lower binding
energies than the parent carbon compounds because of the
higher van der Waals repulsion of the sterically larger
trimethylsilyl moiety. The difference in the binding energies
of very similar compounds can to a first approximation be
equated with the difference in the free energies of binding
(since entropic effects should approximately cancel out), and
the calculated differences in the binding energy DEinter,corr
Scheme 3. Synthesis ofsila-bourgeonal ( 2b).
3368
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3367 –3371

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Angewandte
Chemie
treatment with water and copper(II) chloride[19] then afforded
sila-bourgeonal (2b).
For comparison, sila-lilial (1b) was also prepared by this
route, and the synthesis of sila-bourgeonal (2b) was per-
formed on a multigram scale without purification of the
hydrazone 9 and by employing oxalic acid[20] instead of
copper(II) chloride (see the Supporting Information).
As predicted, all the odorants synthesized have typical
floral-aldehydic lily-of-the-valley odors. Sila-lilial (1b), how-
ever, is somewhat more rosy and fatty in tonality and less
fresh, sparkling, and watery than lilial (1a). The floral, green-
aldehydic fresh-watery lily-of-the-valley note of sila-bourgeo-
nal (2b) lies between that of 1a and 2a, being softer, and less
green-aldehydic than the latter. Therefore, different odor
receptors are certainly involved in their differentiation. Their
floral-aldehydic notes can, however, no longer be discrimi-
nated around their threshold levels. We thus assumed that at
this concentration only the most sensitive lily-of-the-valley
receptor is addressed. The measured odor thresholds
(Table 1) for lilial (1a), bourgeonal (2a), and their silicon
analogues 1b and 2b correlate quite well with the calculated
DEinter,corr values obtained with the hOR17-4 receptor model.
The more negative the enthalpic binding energy DEinter,corr, the
stronger the odorant is bound and the lower the threshold
concentrations required to fire the same amount of recep-
tors.[21] While lilial (1a) and bourgeonal (2a) are comparable
in odor threshold and binding energy, the threshold of sila-
lilial (1b) is about 30 times higher than that of the corre-
sponding carbon compound 1a, which is in good agreement
with the difference of À4.9 kcal in the free binding energy.
The threshold value of sila-bourgeonal (2b) is about four
times higher than that of the carbon compound 2a, which
corresponds to the calculated energy difference of À4.6 kcal.
Even though the correlation is astonishingly good, we
could not take it for granted that the most sensitive lily-of-the-
valley receptor is indeed hOR17-4. It was necessary to
compare the threshold and modeling results with data from
the recombinant and native hOR17-4 receptor. The func-
tional expression of recombinant OR proteins of different
species has already been reported in detail.[22–24] Here we first
expressed hOR17-4 recombinantly in human embryonic
kidney (HEK293) cells. Receptor activation leads singularly
to a transient increase in the cytosolic concentration of
Ca2+ ions which can be detected by ratiometric fluorescence
imaging techniques.[23] The odorants lilial (1a), sila-lilial (1b),
bourgeonal (2a), and sila-bourgeonal (2b), each at 500 mm
concentration, induced transient signals for Ca2+ ions in about
1% of all the cells tested, as is typical for transient olfactory
receptor transfection.[23] In nontransfected HEK cells, signals
for Ca2+ ions were never observed even with a tenfold
increase in the concentration of the agonist. ATP (at a
concentration of 10 mm) served as a control for the excitability
of HEK cells (Figure 2). All four ligands tested gave
Ca2+ signals of nearly the same response amplitude. The
threshold concentration for lilial (1a), sila-lilial (1b), bour-
geonal (2a), and sila-bourgeonal (2b) was found to be in the
5–10 mm range.
Figure 2. Representative Ca2+ imaging recordings ofhOR17-4-trans-
fected HEK293 cells. The cytosolic Ca2+ level of different fura-2-AM-
loaded cells is depicted as the ratio ofthe integrated lfuorescence
(f340/f380) and viewed as a function of time. Compounds 1a, 1b, 2a,
and 2b were applied at a concentration of500 mm for 10 s. ATP
(10 mm) served as a control for the excitability of the HEK cells.
functionally expressed in human spermatozoa and the
spectrum of the effective ligands of human sperms matched
fairly well with the receptive field of the recombinantly
expressed receptors.[9] The spermatozoa are much more
sensitive and exhibited a detection threshold for lilial (1a)
and bourgeonal (2a) at concentrations two orders of magni-
tude lower than those recorded in a HEK cell expression
system.[9] In contrast to HEK293 cells, brief stimulation of
spermatozoa with lilial (1a) and bourgeonal (2a) induced
reproducible and stable Ca2+ signals in a concentration-
dependent manner. Therefore, we compared the potency of
the four ligands 1a, 1b, 2a, and 2b on fura-2-AM-loaded
spermatozoa.
As shown in Figure 3A, all four agonists induced Ca2+
transients in one particular spermatozoon at a concentration
of 50 mm, which guarantees an excellent relative compara-
bility. The amplitudes and signal kinetics of the responses
were similar in magnitude. Coapplication of undecanal, an
aliphatic aldehyde which was demonstrated to be a specific
competitive antagonist for hOR17-4,[9] blocked completely
the response to sila-bourgeonal (2b; Figure 3B) and sila-lilial
(1b), so even if other receptors are expressed in spermatozoa
they would not interfere with this measurement. All four
odorants stimulated spermatozoa in
a dose-dependent
manner. The EC50 values for lilial (1a), sila-lilial (1b),
bourgeonal (2a), and sila-bourgeonal (2b; Table 1) were in
the 10–20 mm range and correspond well with the order of
threshold intensities measured by GC. The threshold concen-
tration for the four agonists of hOR17-4 were in the 100 nm
(1a, 2a, 2b) to 1 mm (1b) range.
Just like there are exceptions in the grammar of a
language, there are many examples of the unpredictability
of odor; some impressive ones were compiled in a recent
minireview by Sell.[25] Just as a native speaker is guided by his
feeling for language, the fragrance chemist will also in the
foreseeable future mainly be led by experience, structural
intuition, imagination, and instinct. In the example of lilial
(1a) and bourgeonal (2a), it was however possible to predict
Only a more qualitative description was possible by using
the HEK293 system. However, the hOR17-4 receptor is also
Angew. Chem. Int. Ed. 2007, 46, 3367 –3371
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3369

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Communications
Experimental Section
The synthesis of 4, the attempt to synthesize 2b by Heck coupling, the
synthesis of 1b by the hydrazone methodology, and the multigram-
scale synthesis of 2b can be found together with general information
in the Supporting Information.
Lilial (1a): Odor: Typical powerful and diffusive aldehydic odor
reminiscent of lily-of-the-valley and linden blossom, mild floral, and
natural. Odor threshold: 0.10 ngLÀ1 air (standard deviation (SD):
0.08).
Bourgeonal (2a): Odor: Powerful and diffusive, watery-floral lily-
of-the-valley note with a green-aldehydic character and hints of
melons and hyacinth. Odor threshold: 0.16 ngLÀ1 air (SD: 0.17).
1b: Pd(OAc)2 (113 mg, 503 mmol) was added to a mixture of 4
(2.77 g, 10.0 mmol), 2-methylprop-2-en-1-ol (865 mg, 12.0 mmol), and
NEt3 (1.21 g, 12.0 mmol). After stirring the reaction mixture for 12 h
at 80–908C, it was allowed to cool to 208C, and the product was
isolated by column chromatography on silica gel (63–200 mm;
pentane/Et2O (9/1)) to afford 1b (1.51 g, 69%) as a colorless liquid.
1H NMR (400.1 MHz): d = 9.71 (d, 3J = 1.5 Hz, 1H; CHO), 7.43 (dXX’
,
4
2H; SiCCH) and 7.14 (dAA’, 2H; CHCCH2, AA’XX’ system, JXX’
=
Figure 3. Single-cell Ca2+ imaging measurements. A representative
ratiometric fluorescence recording of an individual fura-2-AM-loaded
human spermatozoon. The cytosolic Ca2+ level is depicted as the ratio
ofthe integrated lfuorescence ( f340/f380) and viewed as a function of
time. A) Induction oftransient signals ofsimilar amplitude atfer
pulses (10 s) of 1a, 1b, 2a, and 2b at a concentration of50 mm.
B) After preincubation of undecanal for 100 s, the response of sila-
bourgeonal (2b, 50 mm) was completely suppressed by coapplication
(10 s) with undecanal (50 mm).
3
5
4
2
1.5, JAX = 3JA’X’ = 7.6, JAX’ = 5JA’X = 0.6, JAA’ = 1.9 Hz), 3.07 (dd, J =
13.5, 3J = 5.8 Hz, 1H; CH2), 2.72–2.61 (m, 1H; CH), 2.57 (dd, 2J =
13.5, 3J = 8.3 Hz, 1H; CH2), 1.08 (d, 3J = 6.9 Hz, 3H; CHCH3),
0.24 ppm (s, 9H; Si(CH3)3); 13C NMR (100.6 MHz): d = 204.4 (CO),
139.4 (CHCCH2), 138.2 (CSi(CH3)3), 133.6 (2C, SiCCH), 128.4 (2C,
CHCCH2), 48.0 (CHCH3), 36.6 (CH2), 13.3 (CHCH3), À1.1 ppm (3C,
Si(CH3)3); 29Si NMR (79.5 MHz): d = À4.2 ppm; elemental analysis
(%) calcd for C13H20OSi (220.39 gmolÀ1): C 70.85, H 9.15; found: C
70.71, H 9.39. Odor: Lilial-like, typical aldehydic lily-of-the-valley
smell, more rosy and fatty with a slightly spicy connotation, less fresh,
sparkling, and watery than 1a. Odor threshold: 3.30 ngLÀ1 air (SD:
1.77).
2b: A 1.6m solution of BuLi in hexanes (13.8 mL, 22.0 mmol
BuLi) was added dropwise at À788C to a stirred solution of iPr2NH
(2.43 g, 24.0 mmol) in THF (20 mL). The cooling bath was removed
and the mixture allowed to warm to 208C, followed by dropwise
addition of ethanal N,N-dimethylhydrazone[16] (2.07 g, 24.0 mmol) at
À788C. The reaction mixture was again allowed to warm to 208C
(white precipitate), followed by dropwise addition of 8[18] (4.86 g,
20.0 mmol) at À788C with stirring. The reaction mixture was stirred at
208C for 12 h. Subsequently, a saturated aqueous NaCl solution
(20 mL) and pentane (20 mL) were added. The organic layer was
separated and the aqueous layer extracted with pentane (3 ” 20 mL).
The combined organic extracts were dried (MgSO4), the solvent was
removed under reduced pressure, and the residue purified by column
chromatography on silica gel (63–200 mm; pentane/Et2O (1/1)) to
afford 9 (4.73 g, 95%) as a colorless liquid (13C NMR (75.5 MHz): d =
142.1 (CHCCH2), 138.0 (CHN), 137.5 (CSi(CH3)3), 133.5 (2C,
SiCCH), 127.9 (2C, CHCCH2), 43.3 (N(CH3)2), 34.6 and 34.1
(CH2CH2CHN), À1.1 ppm (Si(CH3)3)), which was hydrolyzed. Thus,
a solution of CuCl2·2H2O (7.37 g, 43.2 mmol) in water (36 mL) was
added at 208C to a solution of 9 (4.47 g, 18.0 mmol) in pentane
(180 mL). A few drops of conc. hydrochloric acid were added, and the
mixture was stirred vigorously at 208C for 2 h until the starting
material had been consumed. The layers were separated, the organic
layer was dried (MgSO4) and concentrated under reduced pressure,
and the residue was purified by column chromatography on silica gel
(63–200 mm; pentane/Et2O (5/1)) to afford 2b (2.26 g, 61%) as a
the relative odor intensities of the silicon analogues 1b and 2b
quite accurately on the basis of their stereoelectronic proper-
ties alone from a computational homology model of the
hOR17-4 receptor. At the threshold level, only the recep-
tor(s) of highest affinity should be addressed, and the good
correlation indicates that the hOR17-4 receptor is an
important factor in perception at this threshold concentra-
tion. While the four lily-of-the-valley odorants 1a, 1b, 2a, and
2b can easily be distinguished at higher concentrations by
their additional nuances, they all possess the same floral-
aldehydic note at the threshold concentration. The complex
lily-of-the-valley odor of these odorants above the threshold
level is certainly a result of the activation of different odor
receptors and the mental processing of this information.
However, since the hOR17-4 receptor is activated by these
lily-of-the-valley odorants qualitatively in transfected
HEK293 cells, and quantitatively in the more sensitive
single-spermatozoon model, a corresponding dose–response
relationship could be determined. The different cellular
systems means that the single-cell Ca2+ imaging measure-
ments cannot be compared with the measured GC thresholds
on an absolute basis, but the ranking order is again in
complete agreement. These results taken together clearly
demonstrate that it is indeed the electronic surface structure
that determines the interaction of an odorant with its
olfactory receptors. Thus, the C/Si switching strategy[6] can
even provide insight into the mechanism of receptor activa-
tion in olfaction.
1
3
colorless liquid. H NMR (300.1 MHz): d = 9.81 (t, J = 1.4 Hz, 1H;
CHO), 7.44 (dXX’, 2H; SiCCH) and 7.18 (dAA’, 2H; CHCCH2,
AA’XX’ system, 4JXX’ = 1.5, 3JAX = 3JA’X’ = 7.6, 5JAX’ = 5JA’X = 0.6,
4JAA’ = 1.9 Hz), 2.99–2.89 (m, 2H; CHCCH2), 2.82–2.73 (m, 2H;
CH2CHO), 0.24 ppm (s, 9H; Si(CH3)3); 13C NMR (75.5 MHz): d =
201.5 (CO), 140.9 (CHCCH2), 138.2 (CSi(CH3)3), 133.7 (2C, SiCCH),
127.7 (2C, CHCCH2), 45.1 (CH2CHO), 28.1 (CCH2), À1.1 ppm
(Si(CH3)3); 29Si NMR (59.6 MHz): d = À4.0 ppm; elemental analysis
(%) calcd for C12H18OSi (206.36 gmolÀ1): C 69.84, H 8.79; found: C
3370
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3367 –3371